1327013 - Comparison of Doppler Ultrasound and Self-Gating Techniques for Fetal Cardiac MRI

Congenital heart disease (CHD) occurs in about 0.7-0.9% of births worldwide (1). Despite widespread screening by ultrasound in the United States, up to 65% of cases experience delayed or missed detection (2). Screening by cardiovascular magnetic resonance (CMR) imaging can complement fetal echocardiography (3), but this requires an accurate measure of the fetal heart rate, which is a significant challenge. Three alternative techniques for detecting the fetal heart rate and performing CMR exams are Doppler ultrasound (DUS) gating (4), metric optimized gating (5), and self-gating (SG) methods (6). In this work, we compare the performance of DUS and SG methods in fetal CMR exams.

Methods:

Eight pregnant patients (n=7, 1.5T) and healthy volunteers (n=2, 3T), (age: 30.6 ± 6.2 years, gestational age (GA): 33.1 ± 1.7 weeks) with informed consent underwent successful fetal CMR exams at Children’s Hospital Colorado. One patient (not included) was dropped due to a failure acquiring adequate DUS images. Philips Ingenia 1.5T and 3T scanners were used in the acquisition of axial (AX) and short-axis (SA) orientations of the fetal heart. Breath-holds of 6-10 s (end-expiration) were required for 1-2 slices/breath-hold for a stack of 8-10 slices. The smart-sync DUS device (Northh Medical, Hamburg, DE) was positioned and used to measure the fetal heart rate. Acquisition took place first for the DUS, and then slice locations were matched as close as possible for the SG acquisition. Images were reconstructed to an in-plane resolution of 0.9 – 1.0 mm, with 30 cardiac phases / slice. DUS triggered images were reconstructed inline by the scanner, while SG reconstructions were performed after the exam by setting an appropriate fetal heart rate as determined by principal component analysis of the center point of k-space (7). Full imaging protocol listed in Table 1. Two clinicians scored the images based on a 4-point scale: 1 (poor) borders cannot be traced; 2 (fair) borders are traceable but blurred; 3 (good) borders are slightly blurred; and 4 (excellent) borders are sharp.

Results:

Figure 1 shows example cine images from two subjects at end-systole and end-diastole acquired with DUS and SG. Discrepancies between the image pairs are minor or indistinguishable, despite the inevitability of fetal movement between the acquisitions. Table 2 has the clinicians’ scores for the DUS and SG images.

Conclusion: In principle, DUS had an inherent advantage over the SG techniques because it is measuring the actual fetal heart rate, and any deviations therein are captured automatically. However, in practice these deviations are minimal, and the SG images suffer little, if any, degradation in quality compared to the DUS method. Also, the DUS method may lose signal due to device connectivity, fetal motion, or triggering problems, requiring a repositioning of the device and/or repeat acquisitions. The SG technique does not require any additional hardware, leading to shorter acquisition times and a streamlined workflow.